|
Register | Sign In |
|
QuickSearch
Thread ▼ Details |
Member (Idle past 94 days) Posts: 10333 From: London England Joined: |
|
Thread Info
|
|
|
Author | Topic: The Nature of Scepticism | |||||||||||||||||||||||||||||||||||||||||||
ringo Member (Idle past 441 days) Posts: 20940 From: frozen wasteland Joined:
|
Modulous writes:
The numeric value of the probability is irrelevant. What matters is the risk of being wrong. The boy-who-cried-wolf story is about resource management. Can we afford not to prepare for a real wolf?
Therefore - would it not be true that multiple false positives actually do effect the probability that the next alarm will be correct?
|
|||||||||||||||||||||||||||||||||||||||||||
Stile Member Posts: 4295 From: Ontario, Canada Joined: |
ringo writes: The villagers' "objective" conclusion is based on an inadequate subset of the available evidence. They're ignoring the fact that wolves exist. I never really liked the story either. That's why I didn't include the entire story in my example.I'm making an example in order to simplify things, not complicate them. Bringing the example into the present day, suppose the boy pulls a fire alarm. The firefighters arrive and there's no fire. The boy pulls the alarm again and again the firefighters find no fire. This goes on and on but the firefighters keep coming - because they've drawn an actual objective conclusion that there could be a fire. I agree.The fable and the lesson are kind of silly, as is the example you show here. That's why I simplified the scenario... to take something familiar and delve into some basic aspects. Sorry for any confusion. My issue is that the villagers conclusion is not really objective. Okay. I agree that Aesop was not a scientist? I'm not trying to discuss what the villagers did, or even should do, at all.
|
|||||||||||||||||||||||||||||||||||||||||||
New Cat's Eye Inactive Member |
The numeric value of the probability is irrelevant. What matters is the risk of being wrong. Yeah, sometimes our water pressure drops here at work, and that causes the fire alarm to go off. We have lots of false positives, but we all still grudgingly evacuate the building every time it goes off, even though we're all pretty sure there is no fire.
|
|||||||||||||||||||||||||||||||||||||||||||
AZPaul3 Member Posts: 8563 From: Phoenix Joined: Member Rating: 4.7 |
Conditional probabilities, sir. If the dice reader says it is a six then what is the probability that the dice rolled a six? If your reader is always faithful and is displaying 6 then the outcome of the 1-6 probability came in as a 6
If I told you that my dice reader is designed to never display the correct result ... If your device is a faithful liar then when it says 6 you are assured the roll was not 6. But the roll itself was a 1-6 chance. If the roll landed on 6 then your faithful liar would display a different number.
Your wolf detector has only a 1/365 chance of being right each - and - every - day, regardless of how many times you try. Exactly. Where did this number come from? Did you derive this number from the number of false positives? The 1/365 accuracy rate is indeed established by the number of false positives experienced. Once established that rate, unless you change the detector, remains. I agree.
If so, then you agree that the number of false positives do effect the probability that the next alarm will be correct. There appears to be a vernacular problem here. I'm thinking you are looking at one situation and I am seeing another. Are you saying that after 19 days of false positives you can tentatively say the next day has a 1-20 probability of being right? Then at 39 days of constant false positives you are now saying that the next day has a 1-40 chance of being correct? If so then this is all basakwards. You cannot say anything about the probabilities until you hit on a true positive. I will grant you that on the 39th day you can say that the next day has "at least a 1-40 chance" with the caveat that it may be considerably more than 1-40 and is thus unknown. You have only placed a lower bound.
But can you not see that as the number of false reports goes up, our confidence in the reports goes down? Well, duh. Is this too pedantic?
|
|||||||||||||||||||||||||||||||||||||||||||
Modulous Member Posts: 7801 From: Manchester, UK Joined: |
But can you not see that as the number of false reports goes up, our confidence in the reports goes down? Well, duh. Is this too pedantic? So now are you agreeing with the statement 'multiple false positives actually do effect the probability that the next alarm will be correct'?
|
|||||||||||||||||||||||||||||||||||||||||||
Modulous Member Posts: 7801 From: Manchester, UK Joined: |
The numeric value of the probability is irrelevant. What matters is the risk of being wrong. The boy-who-cried-wolf story is about resource management. Can we afford not to prepare for a real wolf? Largely what I said in Message 236, right?
|
|||||||||||||||||||||||||||||||||||||||||||
AZPaul3 Member Posts: 8563 From: Phoenix Joined: Member Rating: 4.7 |
So now are you agreeing with the statement 'multiple false positives actually do effect the probability that the next alarm will be correct'? At what point? When testing to see how reliable the thing is or once the baseline has been established by having hit a true positive?
|
|||||||||||||||||||||||||||||||||||||||||||
Modulous Member Posts: 7801 From: Manchester, UK Joined: |
At what point? I don't see the need to privilege any particular point.
When testing to see how reliable the thing is or once the baseline has been established by having hit a true positive? In order to have confidence in the true ratio of true to false positives, we'd need several true positives. But if you set a fire alarm in your house that goes off every 10 seconds, you don't need to wait for the day an actual fire occurs to conclude the fire alarm is unreliable and gives off too many false positives. And it would still be the case that an alarm that gives a false positive every 10 seconds is worse than an alarm that gives a false positive every six months.
|
|||||||||||||||||||||||||||||||||||||||||||
AZPaul3 Member Posts: 8563 From: Phoenix Joined: Member Rating: 4.7 |
Agreed. Sorry to have taken your time away from the real issue here. Later.
|
|||||||||||||||||||||||||||||||||||||||||||
RAZD Member (Idle past 1434 days) Posts: 20714 From: the other end of the sidewalk Joined: |
His brother ASSUMED it was a robber and CONCLUDED he was a threat. Sad.
However it is not without flaws. Which is why it is sometimes prudent to error on the side of safety. If one assumes there is always a wolf one will never be surprised when the wolf does come for the herd. Indeed. That was the error the villagers made. If they had been skeptical of their conclusion that should have lead to a different approach. by our ability to understand Rebel American Zen Deist ... to learn ... to think ... to live ... to laugh ... to share. Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click)
|
|||||||||||||||||||||||||||||||||||||||||||
RAZD Member (Idle past 1434 days) Posts: 20714 From: the other end of the sidewalk Joined: |
Then they made a mistake to let him do it again. Indeed, you might say nobody believes a liar...even when he is telling the truth Or you might say the person was unreliable to provide objective evidence to whether or not a wolf was actually around. Next time could be 'true' next time could be 'false' (or not validated), and you could also have a false positive or a false negative.
The point is how a consistent skeptical approach would come to the best solution yes? Yes. So what is the best solution based on your skeptical approach? Give me a few seconds, I'll reveal a simple model in my next post. The past evidence of wolves means one cannot conclude that they are not a threat (the probability of a wolf attack is non-zero). Because the sheep are valuable a more reliable lookout system is needed than one of questionable value - ∴ replace with or add a different person (increase the probability of an accurate report). skeptical of the ability of the boy to provide an accurate reportskeptical of the conclusion that the boy provides false witness by our ability to understand Rebel American Zen Deist ... to learn ... to think ... to live ... to laugh ... to share. Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click)
|
|||||||||||||||||||||||||||||||||||||||||||
RAZD Member (Idle past 1434 days) Posts: 20714 From: the other end of the sidewalk Joined: |
If we personally decide that there is "insufficient evidence" it doesn't mean that what evidence is there no longer exists... it doesn't mean that an objective, evidence-based conclusion cannot be made... it just means that we do not have confidence in that conclusion. Because it is based on opinion/bias/belief ... particularly the assumption that the evidence you have is indicative of reality. The evidence is neither corroborated nor invalidated.
And even if boy may have cried wolf when there was no wolf, you don't have evidence that the next time will be the same. Absolutely true. Conclusions based on objective evidence do not make predictions of the future.
People make predictions of the future. Based on opinion/bias/belief ... and a failure to be skeptical of that conclusion if acted upon.
I'm just trying to show you that objective evidence does still exist, and it's possible that we may personally judge this objective evidence to "not be enough" for us to follow your chart to path B, C or D. Curiously I never said that evidence was non-existent, just that it was not sufficient to form a valid conclusion. One person sighting a wolf in the woods is not sufficient evidence for people to conclude that they are a real threat or that they are not a real threat when nobody else sees it. In this case a decision is necessary to provide adequate protection for the sheep, and several possibilities are available:
Positions a, b, and c would not present a problem, but d would open the possibility of a wolf attack on the sheep occurring, and this would be an undesirable result. Position d means that false positives (c) are possible while waiting for more information but assuming that false negatives (d) do not occur, and that it is okay to climb the mountain for each false or possibly true positive in order to protect the sheep. Position e means thoroughly looking around for wolves within the possible (large) range of wolves. It is reasonable to investigate further due to the potential danger to the sheep. Position f means adding or replacing observers to watch for wolves in order to see if the sighting is replicated while increasing the protection of the sheep due to the potential danger. A cost benefit analysis would likely result in position f being followed. by our ability to understand Rebel American Zen Deist ... to learn ... to think ... to live ... to laugh ... to share. Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click)
|
|||||||||||||||||||||||||||||||||||||||||||
Modulous Member Posts: 7801 From: Manchester, UK Joined: |
Because the sheep are valuable a more reliable lookout system is needed than one of questionable value - ∴ replace with or add a different person (increase the probability of an accurate report). And if a replacement was not possible at the time? Should we always go on wolf alert when the boy cries an alarm, or does there come a time when we should just ignore the boy until such time as a replacement can be arranged?
skeptical of the conclusion that the boy provides false witness What evidence would cause you to change your mind, to cease being sceptical of this conclusion?
|
|||||||||||||||||||||||||||||||||||||||||||
RAZD Member (Idle past 1434 days) Posts: 20714 From: the other end of the sidewalk Joined: |
What evidence would cause you to change your mind, to cease being sceptical of this conclusion? Admission of lying, independent observed real-time invalidation.
And if a replacement was not possible at the time? Should we always go on wolf alert when the boy cries an alarm, or does there come a time when we should just ignore the boy until such time as a replacement can be arranged? Yes, or you might as well not have anyone there. Or you conclude that you don't need those sheep? or that the number that might be taken is an acceptable risk? In which case you don't need anyone there? by our ability to understand Rebel American Zen Deist ... to learn ... to think ... to live ... to laugh ... to share. Join the effort to solve medical problems, AIDS/HIV, Cancer and more with Team EvC! (click)
|
|||||||||||||||||||||||||||||||||||||||||||
Stile Member Posts: 4295 From: Ontario, Canada Joined: |
RAZD writes: Because it is based on opinion/bias/belief ... particularly the assumption that the evidence you have is indicative of reality. The evidence is neither corroborated nor invalidated. This is the issue. This statement can be taken 2 different ways, and they're both wrong.
1. "The evidence being indicative of reality" could be taken to reference the past observation. The objective statement itself:
The boy cries wolf... the village expert wolf-hunters checked and concluded that there was no wolf. Then it's incorrect. This evidence is corroborated and validated... that's what the village experts are for... to corroborate and validate the observation that there is no wolf.Therefore, we have objective, verified information that when the boy cried wolf, there was no wolf (for this one observation, in the past). 2. "The evidence being indicative of reality" could be taken to reference the future observation. The objective conclusion: that based upon this single observation... when the boy cries wolf again, then there will be no wolf at that time. This is, strictly, correct (that this evidence is not indicative of reality). However, it's like trying to show the best path up a mountain and claiming "you have to increase your elevation." No amount of repeated observations can make this "indicative of reality" That's the point, even after many objective, verified observations that the boy cried wolf... the village expert wolf-hunters check and conclude that there was no wolf.... do it 100,000 times in a row. It is still "not indicative of reality" that when the boy cries wolf the next time, then there will be no wolf at that time. There is no such thing as an objective conclusion or theory that is "indicative of reality" for future observations. The only thing that is "indicative of reality" is past observations. Objective conclusions do not make future predictions that are "indicative of reality." The observations don't know the future any more than you do. They only make future predictions that reflect our objective observations from the past. So, what is it?
Later in your recent message, you say something else:
One person sighting a wolf in the woods is not sufficient evidence for people to conclude that they are a real threat or that they are not a real threat when nobody else sees it. This is something we agree on. However, if we did conclude that when the boy cries wolf again, then there will be no wolf... this is not "based on opinion/bias/belief."It is based on objective evidence... the one observation. It's highly likely to be wrong... and there's a very low level of confidence in the conclusion. But it's absolutely wrong to say it's "based on opionion/bias belief" because it's not. It's based on the one objective, verified observation.This makes it an objective conclusion. Not a very confident (good) one... but one all the same. Edited by Stile, : Just some minor clarity.
|
|
|
Do Nothing Button
Copyright 2001-2023 by EvC Forum, All Rights Reserved
Version 4.2
Innovative software from Qwixotic © 2024